BidiMirrored property and ancient scripts (Was Re: Plain text custom fraction input)

Richard Wordingham richard.wordingham at
Sat Jul 25 02:17:07 CDT 2015

On Fri, 24 Jul 2015 23:11:24 +0000
Murray Sargent <murrays at> wrote:

> Richard questions when mirroring is used. As Ken points out, in
> modern BiDi text, such as Arabic and Hebrew, the answer is given by
> the Unicode BiDi Algorithm and associated tables. In ancient scripts
> and in Boustrophedon, it's given by a higher level protocol.

Do you just mean it's determined by the font?  Please give me an actual
example of any other higher level protocol. 

So far as I am aware, in OpenType, anything beyond the Unicode 5.1 Bidi
Mirroring Glyph property actually resides in the font, in features
ltrm, ltra, rtlm and rtla. According to the documentation
these features are applied automatically whenever mirroring appears
to be appropriate for a run.  I'd guess that this means for a resolved
level greater than zero.

Boustrophedon could be given by a higher level protocol.  Are there any
examples of such a higher level protocol?  There are issues with
DIY implementations - text with commas and contour integrals would come
unstuck!  More seriously, I believe there may be a minority of letters
which don't mirror in writing systems where mirroring otherwise happens.

> The UBA wasn't designed to handle mirroring for those scripts.

It wasn't designed for N'ko or Kharoshthi either.  It just happens
to work for them as well.  What is true is that the properties of
Egyptian hieroglyphs weren't designed to work with the UBA.  It is
also true that they weren't set up to work as a script - they're
currently more like mathematical symbols.  As I've noted, there is
work in progress to enable the writing of 'plain-text' Egyptian. 

The UBA's got LRO and RLO.  While not ideal, they ought to work for
scripts where both directions are regularly used. Of course, footnote
numbers would complicate matters, but here we would be getting away
from plain text.


More information about the Unicode mailing list